Is the bad reputation of tantalums justified?

[jitter]

In my work in industrial electronics, I find that the usage of tantalum capacitors is still very common (esp. SMDs), despite having a bad reputation. So I was wondering how justified this reputation actually is.

High capacity and low ESR in a relatively small package are of course attractive characteristics when space is limited. And unlike an electrolytic cap, they don't dry out with age.

In the past, when switching to the RoHS compliant soldering process, we temporarily saw an increase in tantalum failures. And of course the failure mode of a tantalum cap is quite spectacular, probably strengthening the bad reputation even more.

At first we thought that the higher temperature of the leadfree solder process was the culprit. But in all cases an incorrect derating turned out the be the real reason. Look at some tantalum datasheets, and you'll find that the derating can be quite severe. Rule of thumb: use at least twice the rated voltage of the nominal voltage in your circuit. In other words: using a 16 V tantalum on a 12 V rail is asking for problems, use 25 V instead and you're fine.

We do see designs appearing that completely avoid the use of tantalums. Tantalium (Ta) is becoming scarce, and so the prices are rising. That higher price seems to be the actual incentive to avoid tantalums, rather than the alleged short life. And of course, nowadays there are alternatives available (e.g. ceramics of high capacities, polymer caps in the same shapes as SMD-tantalums).

So how justified is that reputation?

[EEVblog]

Tantalums are generally not very tolerant of small overvoltage spikes etc.
When used within spec they are otherwise very reliable.

[jitter]

Might that explain why we saw the most failures during power on? Sometimes as spectecular as shooting a long flame. This destroyed some pcbs, burning right through the FR4 material  .

[John Coloccia]

Hmm, might that explain why we saw the most failures during power on? Sometimes as spectecular as shooting a long flame. This destroyed some pcbs, burning right through the FR4 material  .

Flamethrower is the typical failure mode. Power on surges is a common killer. It's really important to follow all the rules when designing with tantalums. It's not like a lot of other devices where you can get away with a little sloppiness. They are quite serious about things like voltage limits, derating, ripple and things like that. It's like charge pumps. When they say that the maximum PS voltage is 13V, or whatever, you'd better be sure it doesn't see 13.0001V because you'll blow it up. Some devices are just like that.

[Stonent]

I guess I've not been lucky enough to see a spectacular failure of a tantalum.  I have however heard electrolytics go off like a gun shot in cheap computer power supplies. (Complete with paper fluff all over everything, the blue plastic sleeve on the floor outside the power supply and the metal can laying on some random part of the board.

[marshallh]

You really do need to follow every single detail with them, I have seen crowbars around tants and series ferrites.
Otherwise this happens (and this is a very small one)

[AndyC_772]

I've seen them fail short circuit even in designs where they really aren't subject to over-voltage or other stress conditions. Well-respected brand, close to double the required voltage rating, and still we saw a significant number of failed caps.

Ceramics these days are available with such a high capacitance density, I see very little reason to use them any more.

[Smith]

They give a nice flash when they go. Had a dozen of boards running at 12V with 10V caps on it for over 20 years (not my build, and they had no 'obvious' voltage rating). I once hot-plugged one of them looking for a failure, and it went kaboom

Took me days to replace them all.

That was my only bad experience with them in decades, never had any other failing, nor at work, nor at home. I have repaired quite some devices (vintage and new audio, video, laptops etc).

[Psi]

Tantalum on output of Vreg = good  (Stable voltage and their ESR is high enough to not cause stability problems like you get with ceramics)

Tantalum on input of Vreg = bad  (Risk of large input voltage spikes)

[DutchGert]

And dont'forget that the tantalum is mined almost exclusively from conflict countries or at leas from places where they do not care a single bit about polution of the environment.

For me that is reason enough not to use them.

[JoeN]

I put one in backwards in a project once.  100% my fault.  It really did go up like fireworks.  Here is the charring after I removed it (a 10uF cap).

[Joule Thief]

In my experience, they are guaranteed to be a contributor to power supply malfunctions in older 400 series Tek scopes.

[MagicSmoker]

I've seen them fail short circuit even in designs where they really aren't subject to over-voltage or other stress conditions. Well-respected brand, close to double the required voltage rating, and still we saw a significant number of failed caps.

Ceramics these days are available with such a high capacitance density, I see very little reason to use them any more.

This has been my experience with tantalum* caps as well. I have never had a problem with quality aluminum electrolytic capacitors and except for really low profile applications there is no job they - along with a ceramic bypass cap in parallel - can't do better and cheaper than a tant (just use a quality elko, and not one of the notorious Wun Hung Lo brands starring on badcaps.net).


* - more specifically, the "dry sintered" type, rather than the "wet slug" type used almost exclusively for military/space applications; I have no experience with the latter.

[Pjotr]

Tantaliums by themselves are very reliable and long lasting, provided they are properly used. Engineers not understanding and/or neglecting their properties are to be blamed if they fail.

Some limitations are limited overvoltage and limited dI/dt. Limited overvoltage is usually not a problem in a well designed circuit. Usually the cause of faillure is a too high dI/dt, especially when used at power rails. That doesn't mean they are no good at power rails, in fact they perform very well there. But you have to take care to stay within their dI/dt limits, especially at power on this is often neglected.

[IanJ]

Hi,

I used them a LOT in the 80's/90's on offshore installations around the world.....let's say 400 pcb's. All of them were across power rails varying from -15v to +24v and across all of those we probably had less than 5 tantalums go bad (short out....bang!)..............for no apparent reason at the time.

In the workshop during manufacturing design & test though it was a different story.......assemblers putting them in the wrong way round, the wrong voltage etc etc........we probably had 25 short out.....bang!

Never had anything like the same problems with any other passive component.

I can still remember the smell even thought it's been many years!

Ian.

[andybarrett1]

My two pennies worth...

I repair Industrial ABB circuit boards for my day job,,,,,

Rule 1, See a little square Tant cap in a circuit that don't work he be the prime suspect (usually dead short)

Why do ABB love those 6.8uf / 35v things....Pain is the arse to source?

BR
Andy

[MagicSmoker]

Tantaliums by themselves are very reliable and long lasting, provided they are properly used. Engineers not understanding and/or neglecting their properties are to be blamed if they fail.

Or, really? Then how would you explain the ~0.1% failure rate I observed in tantalum caps coming right off the reel, *before* soldering? Can't really blame *me* for that, can you?

Be careful before you call others incompetent.

[Pjotr]

Tantaliums by themselves are very reliable and long lasting, provided they are properly used. Engineers not understanding and/or neglecting their properties are to be blamed if they fail.

Or, really? Then how would you explain the ~0.1% failure rate I observed in tantalum caps coming right off the reel, *before* soldering? Can't really blame *me* for that, can you?

Be careful before you call others incompetent.

Do you feel addressed yourself? With such a high failure rate right from your supplier, complain at you supplier then. It is certainly not common. But yeah, modern Chinese stuff........

Nevertheless the main failure over time is the neglected dI/dt limit of them. This causes a short with fireworks as a final bonus. It is known for over 40 years, but I have seen many designs that did not take care of.

[Howardlong]

I put one in backwards in a project once.  100% my fault.  It really did go up like fireworks.  Here is the charring after I removed it (a 10uF cap).

Easy to do. For some reason, on wire ended caps, they place the band on the +ve not the -ve on tantalums, the opposite of standard electrolytics. I learned to make that mistake just the once.

With polymer electrolytics and ceramics nowadays, there's less and less reason to need tantalums at all.

[SeanB]

My two pennies worth...

I repair Industrial ABB circuit boards for my day job,,,,,

Rule 1, See a little square Tant cap in a circuit that don't work he be the prime suspect (usually dead short)

Why do ABB love those 6.8uf / 35v things....Pain is the arse to source?

BR
Andy

Because that was a standard value, high enough to be useful, but not a 10uF unit which cost twice the price.

I changed many over the years, especially in avionics, mostly the solid slug type which had failed due to them being the cheap ones ( for solid slug that is) with a rubber seal that degraded with time. I used regular aluminium electrolytics as replacements in most cases, just because I could not source the proper ones ( a little sanctions thing), and had a 2 pallet list of everything with the word "cap" in it from stores as a guide to get the stuff out from existing stock. I found a lot of 1000 10uF 10V units in there that had been sitting since the 1970's, and tested 200 before giving up, the only failure was me putting it in backwards and applying 10V with a current limit of 10A. then I used them as small howitzers, especially the ones I removed which still had some capacitance and thus some electolyte.  They do go through thin board at short range though you needed the 28VDC bus rail with the 50A breaker to get the full effect.

The dry ones will go crispy, especially the orange or blue ones, the older coloured band units were a lot more reliable, probably because those have not been made for 40 years so all the early failures would have gone bang by now. The orange would burn through the board, the blue ones would just burn.

I wonder how those ceramic caps will behave with time, I am very suspicious of them, as they are both thin, exposed to contaminants ( no real seal on the active layer) and they have had reliability issues. Ask anybody who used those first gen ceramic decoupling caps in glass tubes, or the yellow dipped cases, and how many caught fire on 5V rails in computers. I have seen them lit up like a lamp on the board, as the power supply was not quite capable of blowing them out, and had the overcurrent limit drop the voltage to around 3V with no foldback.

[MagicSmoker]

Tantaliums by themselves are very reliable and long lasting, provided they are properly used. Engineers not understanding and/or neglecting their properties are to be blamed if they fail.

Do you feel addressed yourself? With such a high failure rate right from your supplier, complain at you supplier then. It is certainly not common. But yeah, modern Chinese stuff........

You made the blanket statement bolded/underlined above and so I was deemed incompetent by you via implication.

That said, the capacitors which were DOA on the reel were made by AVX, so not a "no-name" Chinese supplier (although I am sure they were made in China, just like every other electronic component these days), however, identical value/package capacitors from Kemet and Cornell-Dubilier also failed in circuit.

Nevertheless the main failure over time is the neglected dI/dt limit of them. This causes a short with fireworks as a final bonus. It is known for over 40 years, but I have seen many designs that did not take care of.

Yes, the dI/dt limit for tantalum caps - as well as failures due to ripple and surge currents, and overvoltage - are reasonably well known, especially among those who design or service switch-mode converters and limiting surge current into or out of the capacitors is easy enough to do. I am saying that even when I have addressed all of the aforementioned restrictions I still had tantalum caps fail prematurely in-circuit (as well as straight off the reel).

[jitter]

(although I am sure they were made in China, just like every other electronic component these days)

Or so we think.

Some examples:
Look at ON-Semi devices with a lot code that starts with R. That's for Roznov in the Czech Republic.
The other day I was looking at a reel of IM03GR relays from TE/Axicom: made in Portugal. Same brand, different type: Czech Republic.
OP627AP's: Thailand. The list could be continued with: Malaysia, Philippines, Israel (e.g. Vishay/Kemet), Korea etc.

[G0HZU]

Tantaliums by themselves are very reliable and long lasting, provided they are properly used. Engineers not understanding and/or neglecting their properties are to be blamed if they fail.

That's my experience too. I've designed large boards with lots of SMD tants on. i.e. dozens and dozens of them per PCB and we made many, many thousands of these boards with no reliability issues at all. I've been using tants in all my designs for over 25 years and if I thought there was an issue with them then I'd stop using them and so would the company I work for. So I speak for over 100 engineers across maybe 30 years of designing high end comms products. The bottom line is that tants are high performance components with excellent reliability over time and temperature etc provided they are used correctly

I've seen what happens if an engineer uses them incorrectly. Usually the failure rate becomes quite high with many tants failing during initial development or during initial acceptance testing. In a bad design they can fail spectacularly the first time the unit is switched on.

[sarepairman2]

i always thought tantalums is something you put on the pampered isolated front end that uses a bunch of linear regulators.

i wonder if alot of failures are due to spikes on power on...

perhaps i should start adding littleRC filters to the en pin on regulators.

[MagicSmoker]

...So I speak for over 100 engineers across maybe 30 years of designing high end comms products. The bottom line is that tants are high performance components with excellent reliability over time and temperature etc provided they are used correctly
...
In a bad design they can fail spectacularly the first time the unit is switched on.

Okay, so how do you use them correctly, or was the purpose of your post - like Pjotr's earlier - merely to tell us how awesome you are?